Getting started

Installation

Requirements

  • python 3

  • numpy

  • networkx

  • pandas

  • scipy

  • scikit-learn

  • statsmodels

  • pydot

(For visualization)

  • matplotlib

  • graphviz

  • pygraphviz (might not support the most recent Mac)

Install via PyPI

To use causal-learn, we could install it using pip:

(.venv) $ pip install causal-learn

Install from source

For development version, please kindly refer to our GitHub Repository.

Running examples

For search methods in causal discovery, there are various running examples in the ‘tests’ directory in our GitHub Repository, such as TestPC.py and TestGES.py.

For the implemented modules, such as (conditional) independent test methods, we provide unit tests for the convenience of developing your own methods.

Quick benchmarking

To help users get a quick sense of the running time of the algorithms of interest, we conducted benchmarking for several methods. We consider datasets with number of variables from {10, 25, 50, 100} and average degree from {2, 3, 4, 5}. The random graphs are Erdős-Rényi graphs. The average degree is the average number of edges connected to a node. We simply calculate it by dividing the sum of degrees by the total number of nodes in the graph. The sample size is 1000. All algorithms were run on a single cluster node with 8 CPUs (Intel Xeon E5-2470) and 16 GB Memory. We denote running time as ‘>D’ if it is more than one day. All results are average over 10 runs with different random seeds.

_images/table_modified.png

Contributors

Team Leaders: Kun Zhang, Joseph Ramsey, Mingming Gong, Ruichu Cai, Shohei Shimizu, Peter Spirtes, Clark Glymour

Coordinators: Biwei Huang, Yujia Zheng, Wei Chen

Developers: